Ebook viewer: handle self closed tags in XHTML docs correctly. Fixes #3727 (Ebook-viewer problems with valid xhtml)

src/calibre/ebooks/metadata/opf2.py

@@ -270,6 +270,7 @@ class Spine(ResourceCollection):
             Resource.__init__(self, *args, **kwargs)
             self.is_linear = True
             self.id = idfunc(self.path)
+            self.idref = None
 
     @staticmethod
     def from_opf_spine_element(itemrefs, manifest):
@@ -281,6 +282,7 @@ class Spine(ResourceCollection):
                 if path:
                     r = Spine.Item(s.manifest.id_for_path, path, is_path=True)
                     r.is_linear = itemref.get('linear', 'yes') == 'yes'
+                    r.idref = idref
                     s.append(r)
         return s
 

src/calibre/ebooks/oeb/iterator.py

@@ -19,6 +19,7 @@ from calibre.utils.zipfile import safe_replace, ZipFile
 from calibre.utils.config import DynamicConfig
 from calibre.utils.logging import Log
 from calibre.ebooks.epub.output import EPUBOutput
+from calibre import guess_type
 
 TITLEPAGE = EPUBOutput.TITLEPAGE_COVER.decode('utf-8')
 
@@ -39,20 +40,20 @@ class UnsupportedFormatError(Exception):
 
 class SpineItem(unicode):
 
-    def __new__(cls, *args):
+    def __new__(cls, path, mime_type=None):
-        args = list(args)
-        path = args[0]
         ppath = path.partition('#')[0]
         if not os.path.exists(path) and os.path.exists(ppath):
             path = ppath
-        args[0] = path
+        obj = super(SpineItem, cls).__new__(cls, path)
-        obj = super(SpineItem, cls).__new__(cls, *args)
         raw = open(path, 'rb').read()
         raw, obj.encoding = xml_to_unicode(raw)
         obj.character_count = character_count(raw)
         obj.start_page = -1
         obj.pages      = -1
         obj.max_page   = -1
+        if mime_type is None:
+            mime_type = guess_type(obj)[0]
+        obj.mime_type = mime_type
         return obj
 
 class FakeOpts(object):
@@ -150,8 +151,17 @@ class EbookIterator(object):
         self.language = self.opf.language
         if self.language:
             self.language = self.language.lower()
-        self.spine = [SpineItem(i.path) for i in self.opf.spine if i.is_linear]
+        ordered = [i for i in self.opf.spine if i.is_linear] + \
-        self.spine += [SpineItem(i.path) for i in self.opf.spine if not i.is_linear]
+                  [i for i in self.opf.spine if not i.is_linear]
+        self.spine = []
+        for i in ordered:
+            spath = i.path
+            mt = None
+            if i.idref is not None:
+                mt = self.opf.manifest.type_for_id(i.idref)
+            if mt is None:
+                mt = guess_type(spath)[0]
+            self.spine.append(SpineItem(spath, mime_type=mt))
 
         cover = self.opf.cover
         if self.ebook_ext in ('lit', 'mobi', 'prc', 'opf') and cover:

src/calibre/gui2/viewer/documentview.py

@@ -18,7 +18,7 @@ from calibre.gui2.viewer.config_ui import Ui_Dialog
 from calibre.gui2.viewer.js import bookmarks, referencing, hyphenation
 from calibre.ptempfile import PersistentTemporaryFile
 from calibre.constants import iswindows
-from calibre import prints
+from calibre import prints, guess_type
 
 def load_builtin_fonts():
     base = P('fonts/liberation/*.ttf')
@@ -352,6 +352,8 @@ class DocumentView(QWebView):
     def __init__(self, *args):
         QWidget.__init__(self, *args)
         self.debug_javascript = False
+        self.self_closing_pat = re.compile(r'<([a-z]+)\s+([^>]+)/>',
+                re.IGNORECASE)
         self.setSizePolicy(QSizePolicy(QSizePolicy.Expanding, QSizePolicy.Expanding))
         self._size_hint = QSize(510, 680)
         self.initial_pos = 0.0
@@ -447,8 +449,14 @@ class DocumentView(QWebView):
 
     def load_path(self, path, pos=0.0):
         self.initial_pos = pos
+        mt = getattr(path, 'mime_type', None)
+        if mt is None:
+            mt = guess_type(path)[0]
         html = open(path, 'rb').read().decode(path.encoding, 'replace')
         html = EntityDeclarationProcessor(html).processed_html
+        if 'xhtml' in mt:
+            html = self.self_closing_pat.sub(r'<\1 \2></\1>', html)
+        #self.setContent(QByteArray(html.encode(path.encoding)), mt, QUrl.fromLocalFile(path))
         self.setHtml(html, QUrl.fromLocalFile(path))
         self.turn_off_internal_scrollbars()